parse_reorg(options, params)

l = parse_reorg(options, params);

layer parse_reorg(list *options, size_params params)
{
	//设置参数
    int stride = option_find_int(options, "stride",1);
    int reverse = option_find_int_quiet(options, "reverse",0);
    int flatten = option_find_int_quiet(options, "flatten",0);
    int extra = option_find_int_quiet(options, "extra",0);

    int batch,h,w,c;
    h = params.h;
    w = params.w;
    c = params.c;
    batch=params.batch;
    if(!(h && w && c)) error("Layer before reorg layer must output image.");
	//设置网络层
    layer layer = make_reorg_layer(batch,w,h,c,stride,reverse, flatten, extra);
    return layer;
}

layer layer = make_reorg_layer(batch,w,h,c,stride,reverse, flatten, extra);

layer make_reorg_layer(int batch, int w, int h, int c, int stride, int reverse, int flatten, int extra)
{
    layer l = {0};
    l.type = REORG;
    l.batch = batch;
    l.stride = stride;
    l.extra = extra;
    l.h = h;
    l.w = w;
    l.c = c;
    l.flatten = flatten;
    //根据reverse获取输出特征图尺寸和通道
    if(reverse){
        l.out_w = w*stride;
        l.out_h = h*stride;
        l.out_c = c/(stride*stride);
    }else{
        l.out_w = w/stride;
        l.out_h = h/stride;
        l.out_c = c*(stride*stride);
    }
    l.reverse = reverse;
	//输出大小
    l.outputs = l.out_h * l.out_w * l.out_c;
    l.inputs = h*w*c;
    if(l.extra){
        l.out_w = l.out_h = l.out_c = 0;
        l.outputs = l.inputs + l.extra;
    }

    if(extra){
        fprintf(stderr, "reorg              %4d   ->  %4d\n",  l.inputs, l.outputs);
    } else {
        fprintf(stderr, "reorg              /%2d  %4d x%4d x%4d   ->  %4d x%4d x%4d\n",  stride, w, h, c, l.out_w, l.out_h, l.out_c);
    }
    //一个batch输出大小
    int output_size = l.outputs * batch;
    l.output =  calloc(output_size, sizeof(float));
    l.delta =   calloc(output_size, sizeof(float));

    l.forward = forward_reorg_layer;//前向网络层
    l.backward = backward_reorg_layer;//反向传播
#ifdef GPU
    l.forward_gpu = forward_reorg_layer_gpu;
    l.backward_gpu = backward_reorg_layer_gpu;

    l.output_gpu  = cuda_make_array(l.output, output_size);
    l.delta_gpu   = cuda_make_array(l.delta, output_size);
#endif
    return l;
}

l.forward = forward_reorg_layer;//前向网络层

void forward_reorg_layer(const layer l, network net)
{
    int i;
    if(l.flatten){
        memcpy(l.output, net.input, l.outputs*l.batch*sizeof(float));
        if(l.reverse){
            flatten(l.output, l.w*l.h, l.c, l.batch, 0);
        }else{
            flatten(l.output, l.w*l.h, l.c, l.batch, 1);
        }
    } else if (l.extra) {
        for(i = 0; i < l.batch; ++i){
            copy_cpu(l.inputs, net.input + i*l.inputs, 1, l.output + i*l.outputs, 1);
        }
    } else if (l.reverse){
        reorg_cpu(net.input, l.w, l.h, l.c, l.batch, l.stride, 1, l.output);
    } else {//执行此步
    
        reorg_cpu(net.input, l.w, l.h, l.c, l.batch, l.stride, 0, l.output);
    }
}
//具体可查看博客https://blog.csdn.net/qq_17550379/article/details/78948839
void reorg_cpu(float *x, int w, int h, int c, int batch, int stride, int forward, float *out)
{
    int b,i,j,k;
    int out_c = c/(stride*stride);

    for(b = 0; b < batch; ++b){
        for(k = 0; k < c; ++k){
            for(j = 0; j < h; ++j){
                for(i = 0; i < w; ++i){
                    int in_index  = i + w*(j + h*(k + c*b));
                    int c2 = k % out_c;
                    int offset = k / out_c;
                    int w2 = i*stride + offset % stride;
                    int h2 = j*stride + offset / stride;
                    int out_index = w2 + w*stride*(h2 + h*stride*(c2 + out_c*b));
                    if(forward) out[out_index] = x[in_index];
                    else out[in_index] = x[out_index];
                }
            }
        }
    }
}

void backward_reorg_layer(const layer l, network net)
{
    int i;
    if(l.flatten){
        memcpy(net.delta, l.delta, l.outputs*l.batch*sizeof(float));
        if(l.reverse){
            flatten(net.delta, l.w*l.h, l.c, l.batch, 1);
        }else{
            flatten(net.delta, l.w*l.h, l.c, l.batch, 0);
        }
    } else if(l.reverse){
        reorg_cpu(l.delta, l.w, l.h, l.c, l.batch, l.stride, 0, net.delta);
    } else if (l.extra) {
        for(i = 0; i < l.batch; ++i){
            copy_cpu(l.inputs, l.delta + i*l.outputs, 1, net.delta + i*l.inputs, 1);
        }
    }else{
        reorg_cpu(l.delta, l.w, l.h, l.c, l.batch, l.stride, 1, net.delta);
    }
}